Skip to main content

Large-scale AM and construction

📍 Where we are: Part IV · Materials and New Systems — Chapter 11 of 19. The previous chapter ended at the micrometre scale; this one goes the other way, to parts the size of boat hulls, bridges and walls, and closes Part IV before the industrial chapters.

The simple version

Icing a cupcake is easy. Icing a wedding cake taller than you, with the bottom layers still soft while you stack more on top, is a different craft. Scaling up additive manufacturing is like that: the same principle, but heat, weight and time now work against you. Each layer cools while the next goes on, the material at the bottom must carry what is above it, and a mistake at hour three shows up at hour thirty.

What this chapter covers​

Three large-scale families: wire-arc additive manufacturing (WAAM) of metal, large-format polymer printing with pellet extruders, and 3D construction printing of concrete. For each we ask what has been shown in practice, what remains lab-scale, and what is contested. The common theme is that process stability, tolerance and heat or moisture management become the central problems once parts are big.

Problem​

Powder bed machines are limited by build chamber size, and the cost of the machine and powder rises quickly with volume. Large parts such as moulds, tooling, structural nodes and building elements need high deposition rates and low-cost feedstock: wire for metal, pellets for polymer, cementitious mortar for buildings. The price is control. The deposited volumes are large, process time is long, and thermal and rheological history accumulate in ways a small build never sees. Where the part is a structure that carries people, the added question is whether any code or standard recognises it.

State of the art​

Wire-arc additive manufacturing (WAAM)​

WAAM deposits metal wire with an arc heat source, a route to large metal components at relatively high deposition rates and low material cost [1][2]. The best-known structure is the MX3D stainless-steel footbridge. Gardner et al. report testing and initial verification of the 10.5 m span bridge: tensile coupon tests for the material, stub column tests for key components, full-scale load tests during and after construction, and finite-element comparison; the bridge carried its full serviceability-limit-state design load, which allowed it to be shown to the public with controlled access at Dutch Design Week 2018, with further testing and modelling planned before final installation [3]. Evidence level: a demonstrated one-off structure with a published verification study, not yet a codified product.

The recurring defects in WAAM are residual stress and distortion, cracks, porosity and humping; their causes include wire-feed-to-travel-speed ratio, heat input, contact-tip-to-work distance and gas flow [4]; an earlier review likewise lists deformation, porosity and cracking as the common defects across alloys [2]. MX3D, which builds metre-scale parts, explains that depositing hundreds of kilograms over several days can cause severe distortion, warping or delamination if unmanaged, and addresses this through path planning with alternating directions, symmetrical balance and skip sequences; for titanium, low thermal expansion limits distortion but very low thermal conductivity requires precise monitoring to prevent local heat build-up [5]. That is a vendor engineering guide, so we treat it as practitioner knowledge, not as a measured comparison.

Large-format polymer printing​

Pellet-fed large-format additive manufacturing (LFAM), such as Oak Ridge National Laboratory's big-area additive manufacturing (BAAM), has higher deposition rates and lower material cost than filament-based extrusion, and better properties with carbon or glass fibre [6]. ORNL's early technical review describes a 7.5 mm extrusion diameter and deposition above 5,000 cm³ per hour, and explains that temperature gradients produce residual stress seen as warp and curl; carbon fibre cuts the coefficient of thermal expansion sharply but directionally, with 13% carbon fibre in ABS giving 106 µm/m/°C perpendicular to the bead and as low as 9.8 along it [7].

A 2023 review adds the other side of fibre reinforcement: the fibres lie in plane, so none span from layer to layer, and the proposed mitigation "z-pinning" places voids across layers to be filled in one deposition; it also describes ORNL gantry-mounted thermal-camera feedback on layer time, and case studies such as an 11 m boat mould [8]. A 2024 review covers interlayer bonding, surface quality against speed, and recyclability [9]. Early characterisation of Oak Ridge's Big Area Additive Manufacturing (BAAM) material found that voids and imperfect fusion between layers can substantially reduce strength, with failure at bead and layer interfaces [10]; a review of interlayer bonding in fused filament fabrication documents the mechanisms and enhancement methods for that weak interface [11]. Evidence level: production for moulds and tooling, lab-scale for property-recovery methods.

3D-printed concrete​

Bos et al. described the field in 2016 as being in its infancy and presented the Eindhoven 3D Concrete Printing facility, a 4-degree-of-freedom gantry robot with a 9 × 4.5 × 2.8 m print area in operation since September 2015 [12]. A 2018 research roadmap classified the fresh and hardened material properties that govern printed geometry [13], and since then the technical issues have been defined more sharply.

  • Interlayer bond. A review of concrete 3D-printed structural members states that a weak interlayer bond creates a weak plane without reinforcement and gives anisotropic behaviour [14], and a critical review of the interlayer region discusses the parameters that control bond strength and strategies such as mechanical interlocking and surface or interface tailoring [15]. Experimental work on hardened properties showed that process parameters influence interlayer adhesion [16].
  • Reinforcement. This is the field's structural hole. The same 2022 review says integration of reinforcement during printing remains a challenge and that no robust, efficient solution had been proposed [14]. Fibre-reinforced printable concrete is one route; steel fibres improved strength more than they improved interlayer bond, compared with carbon or glass fibres [17]. A 2025 review finds steel bars combined with fibres and nanomaterials remain a key strategy [18]. Two experimental studies tested alternatives: entraining a metal cable in the concrete filament during printing, with bond behaviour characterised [19], and four-point bending tests of printed beams with various reinforcement types [20].
  • Codes and standards. ISO/ASTM 52939:2023 specifies quality-assurance requirements for additive construction in building and construction projects; it is independent of material and process category but does not apply to metals [21]. We did not read the full standard, which is paywalled. In the United States, the International Code Council reports that ICC 1150, the Standard for Automated Construction Technology for 3D Printing Walls, became ANSI approved on 15 July 2026; it sets baseline requirements for materials, qualification, inspection and structural design of cementitious 3D-printed walls, load-bearing or not, with or without steel reinforcement, with design coverage for gravity, wind and a restricted level of seismic loading [22]. We rely on the ICC's own journal article for this status and have not read the standard text. A NIST workshop report notes a lack of field data hinders refinement of guidelines [23].

Evidence level: demonstrated for printed walls and non-structural elements, where permitting is done case by case; contested for structural reinforcement strategies.

Why it is hard​

  1. Thermal history at scale. Metal and polymer parts accumulate heat, and distortion grows with deposited mass and time. The cure is process planning, not a single parameter.
  2. Process stability over days. A deposition run of many hours must hold dimensions and quality throughout, and a late failure scraps a large amount of material. Monitoring and control (In-situ monitoring and metrology, Closed-loop process control) are much less mature for arc, pellet and concrete deposition than for powder bed fusion.
  3. Anisotropy. Layers are not equal in strength to the bulk, for fibre polymer in z, and for concrete across layers (Microstructure, anisotropy and process-structure-property links).
  4. Tolerance. Big parts, rough as-deposited surfaces and shrinkage or distortion mean machining or finishing stays in the loop.
  5. Reinforcement and codes. Concrete is weak in tension, and the construction industry's trust is built on codes that were written for cast or formed elements.

Three large-scale processes with their dominant problem: WAAM with distortion and residual stress, pellet extrusion with warp and z-weakness, concrete printing with interlayer bond and reinforcement; with the problems common to all three below. Large-scale AM families side by side: feedstock, dominant problem and evidence level. Original diagram by the authors, created with AI assistance.

Open questions​

  • Can reinforcement be integrated into concrete printing, by rebar, fibres, mesh or hybrid methods, in a way that is automated, certifiable and does not limit geometry?
  • Can in-process sensing and control for WAAM and pellet deposition reach the maturity of powder bed monitoring?
  • What field data, and what test methods, are needed to turn ISO/ASTM 52939 qualification principles into design codes? This is largely contested for structural use.
  • How should the anisotropy of large-format polymer and concrete parts be characterised and allowed for in design?
  • How much finishing and machining is acceptable before the cost advantage disappears? (Cost, throughput and scaling to production)

Industry impact​

Large-scale AM is mostly about cost and lead time for tooling, moulds, building elements and large structures where powder-bed machines cannot go. Pellet printing for moulds is the most mature, WAAM has a landmark structure and growing use in large metal components, and construction printing has real buildings and a standard for qualification principles, with load-bearing reinforcement the open frontier. The industrial chapters that follow (Qualification and certification) apply to all of these, and the construction case is a live example of what happens when a technology precedes its codes.

Why it matters​

The big-scale applications are where AM meets the oldest, most conservative industries. Their adoption depends less on spectacular demonstrations than on dull items: field data, codes, finish and a controlled process that stays stable for a hundred hours.

In the real world​

The MX3D bridge, ORNL's BAAM moulds such as the 11 m boat mould, and the TU Eindhoven concrete printer are the cited public examples; the ISO/ASTM 52939 standard and ICC 1150 are the regulatory context for building elements.

Key terms​

  • Wire-arc additive manufacturing (WAAM) — depositing metal wire melted by an electric arc, layer by layer.
  • Large-format additive manufacturing (LFAM) — printing parts at metre scale, typically with pellet-fed extruders.
  • Big-area additive manufacturing (BAAM) — Oak Ridge National Laboratory's pellet-extrusion large-format system.
  • 3D concrete printing (3DCP) — extruding cementitious mortar in layers to build structures.
  • Interlayer bond — the adhesion between successive deposited layers.
  • Z-pinning — bridging layers with material deposited across them to counter weakness in the build direction.
  • Humping — periodic bead irregularity in arc deposition.
  • Coefficient of thermal expansion (CTE) — how much a material grows per degree of temperature rise.
  • ISO/ASTM 52939 — the 2023 qualification-principles standard for additive manufacturing in construction.
  • Rebar — steel reinforcing bar placed in concrete to carry tension.

Where this leads​

The industrial chapters begin with the bottleneck that touches everything above: Qualification and certification: the central bottleneck.